Tracking the Burden Surface Radial Profile of a Blast Furnace by a B-mode Mechanical Swing Radar System

抄録

To continuously track the burden surface radial profile inside a blast furnace in every noncharging period, a new organizing structure for the scanned data of a mechanical swing radar system was proposed. The detection results of the radial shape of the burden surface in one scanning period can be organized into a matrix and represented by a composite image. Then, the extraction of the burden surface radial profile can be achieved by the segmentation of a featured region in the composite image. To address the incorrect segmentation results caused by the deterioration of the image quality in the later stage of each noncharging period, a priori curve-based image segmentation algorithm was proposed. The shape prior was constructed by a priori shape function and a current state function decomposed from the contours of the priori and current segmented regions, respectively. Compared with the classical region-scalable fitting segmentation algorithm, the proposed algorithm has the ability to provide more reasonable segmentation results during the entire noncharging period. The tracking of the burden surface radial profile can be accomplished by calculating the corresponding shape function from the contour of the segmented image region. Compared with the results produced by the existing A-mode radar data processing method, oscillations and local outliers can be avoided in the results of the proposed method. The goal of the continuous tracking of the burden surface radial profile was accomplished.